Does Lightroom's Open Eyes Feature Actually Look Natural?

Lightroom can now open closed eyes in a photo, a job that used to mean digging through a burst for a donor frame and blending it in by hand in Photoshop. If you shoot portraits, events, or group photos, a blink is one of the most common reasons a frame with good expression and good light lands in the reject pile.

The feature is called Open Eyes, and for now it lives only in the Lightroom mobile app. Aaron Nace of PHLEARN put it to the test on his own face, using a handful of frames from a recent tutorial shoot where he'd been caught mid-blink. Once Lightroom analyzes a photo and spots closed eyes, a new Open Eyes icon appears at the bottom of the screen. The generation doesn't run on Firefly. Adobe routes it through partner models instead, and in the video, a label for Google's Nano Banana 2 flashes on screen near the end of processing. Nace's first attempt cost 20 generative credits, and his short answer on whether the tool works is "kind of."

Before it changes any pixels, Lightroom looks for other photos from the same shoot and uses them as reference, reading both the eyes and the lighting in the scene so the new eyes have a better chance of fitting. On the first generation, Nace's eyes opened, the rest of his face stayed put, and his eyebrows lifted enough to make him look mildly surprised. Zoomed in next to a frame where his eyes were really open, the catchlight held up well and the color came close without landing exactly. If a result misses, you reset the photo and generate again, and every new attempt costs another round of credits.

Google's Best Take, introduced with the Pixel 8 in 2023, tackled the same problem for group shots by pulling faces from a burst of nearby frames and letting you swap in the best expression for each person. Retouchers have done a manual version of that for years with head swaps in Photoshop. Lightroom's approach sits somewhere in between: it borrows from your other frames as reference, then generates new eyes rather than copying existing pixels. Either way, the habit that pays off is the same one wedding and event shooters already rely on, which is firing several frames of any group or portrait moment instead of one. Extra frames give the software more to match against, and they also give you the option of skipping the AI entirely if one of those frames has everyone's eyes open. For a crowded group shot where one guest blinked, that combination of a real backup frame and a one-tap fix is a lot faster than building a composite from scratch.

When Nace loaded a larger 24-megapixel file, the same operation used 40 credits instead of 20, so higher-resolution files from a modern camera will burn through your monthly allotment faster than phone snapshots. Adding more images from the same session also changed the process itself, with Lightroom inserting an extra step labeled as matching the subject's eyes across the set. His second generation opened the eyes without moving the eyebrows at all.

The side-by-side of his real eyes and both AI versions is in the video above.